Locus-Level Transposable Element Profiling Resolves Division-Coupled Transcriptional Dynamics During Human Endoderm Specification

Abstract

BackgroundTransposable elements (TEs) constitute nearly half of the human genome and are now recognized as significant contributors to mammalian gene regulatory networks. Despite this, most transcriptomic studies quantify TE expression at the subfamily level, which may obscure meaningful variation arising from individual insertion sites. Whether resolving TE expression to individual loci can reveal biologically distinct signals during stem cell differentiation has not been systematically characterized. ResultsWe re-analysed a published RNA-seq time course of FUCCI-h9 human embryonic stem cells differentiating into definitive endoderm (0-72 hour, seven time points, three division cycles, three biological replicates), quantifying expression in parallel at two complementary resolutions: TE subfamilies using TEtranscripts and individual TE loci using TElocal. The primary finding is that individual TE loci capture heterogeneous transcriptional responses at cell division boundaries that are entirely absent at the subfamily level. Within-division-state PC1 variance for TE loci was substantially elevated at the first division cycle (6.72; 95% bootstrap CI 0.63-8.48) compared with TE subfamilies (0.22; CI 0.04-0.34), with non-overlapping confidence intervals providing statistically robust support for the resolution advantage. Differential expression analysis identified over 18,000 dynamic TE loci across the time course, exceeding the 268 differentially expressed subfamilies, with alternating phases of silencing and reactivation resolved only at locus resolution. Differentially expressed TE loci were non-randomly enriched at superenhancers (fold enrichment: 9.1-23.5-fold; p<0.001 by permutation), with peak overlap at 36-48 hours coinciding with the second cell division and endoderm commitment. ERV1-class elements, particularly HERVH-int, were the dominant contributors, and representative loci near the endoderm regulators MIXL1 and ID3 showed differentiation-induced RNA-seq signal within proximal superenhancer domains. ConclusionsTE loci exhibit heterogeneous transcriptional responses at cell division boundaries, a signal with non-overlapping bootstrap confidence intervals relative to TE subfamilies at the first division cycle that is entirely masked by subfamily-level aggregation. This division-boundary resolution advantage, together with permutation-confirmed enrichment of dynamic loci at superenhancers during a discrete 36-48 hour endoderm commitment window, supports broader adoption of locus-resolved TE quantification as a complement to conventional gene expression analysis in developmental genomics.